Have a UROP opening you would like to submit?
Please fill out the form.
Physically Explainable Deep Learning for Chemical Systems
2: Mechanical Engineering
In the last decades, deep learning has achieved enormous progress in various tasks in computer science, including computer vision and natural language processing. We want to explore the potential of deep learning in learning the interactions among high-dimensional species during chemical energy conversion, especially the combustion process. Our recent work has formulated a framework to build physically explainable neural networks to discover chemical reaction pathways from data, termed as chemical reaction neural network (CRNN). We plan to design a general tool for learning reaction pathways for various chemical systems, including combustion, battery, and biology. The main goal of this project is to improve the learning algorithms and software infrastructure of CRNN. Additionally, you will help improve the clarity of the code and documentations in GitHub repository, as we plan to release the code to the research community.
Experience with Python/Julia and machine learning is required. Experience with deep learning in Tensorflow/Pytorch/Julia is preferred but not required. Exposure to differential equations, optimization, and numerical simulation is helpful but not required.